Querying XML Data

1
Querying XML Data

• By Willie Milnor
• LSDIS Lab
• Computer Science Department
• University of Georgia

2
Data Storage

• Relational databases
• relations/tables of records
• attributes
• records are connected via foreign keys
• XML
• hierarchical structure of elements
• nested attributes or sub elements
• elements connected via references to element Ids
  (IDREFs)

3
Simple Example

• Relational database
• author
• au1 Matthew Gambardella Atlanta, GA
• 770-350-9955
• book
• bk15 XML Developer's Guide Computer 44.95
• 2000-10-01 au1

4
Simple Example

• XML
• ltauthor idau1gt
• ltnamegt Matthew Gambardella lt/namegt
• ltaddressgt Atlanta, GA lt/addressgt
• ltphonegt 770-350-9655 lt/phonegt
• lt/authorgt
• ltbook id"bk15 authorau1gt
• lttitlegtXML Developer's Guidelt/titlegt
  ltgenregtComputerlt/genregt
• ltpricegt44.95lt/pricegt
• ltpublish_dategt2000-10-01lt/publish_dategt
• lt/bookgt

5
Schemas

• Structure of data represented in schemas
• Relational databases use a DDL
• XML
• DTD
• XML Schema
• Specific schema langauages for XML-like languages
• i.e. RDF/S for RDF
• no schema
• self-describing data

6
Why Use Semi-Structured Data

• Raw data is often semi-structured
• script generated
• HTML has inherent semi-structure
• Less rigid structure than traditional databases
• may not have predefined schema
• convenient

7
Why Use Semi-Structured Data

• Integration of heterogeneous data sets
• schemas do not enforce typing
• single-/multi-valued attributes
• Standardize information exchange
• XML is the de facto format
• RDF in Semantic Web
• Ability to browse/query without full knowledge of
  the schema

8
Why Use XML

• XML (Extended Markup Language) has emerged as a
  standard for exchanging data on the Web
• XML can be used on a wide variety of platforms
  and interpreted with a wide variety of tools
• many available parsers
• Enables separation of content (XML) and
  presentation
• XSL (Extensible Stylesheet Language)

9
Challenges Posed by XML

• extract data from large XML documents
• exchange (export) XML data
• exchange XML among different, related ontologies
• integrate XML from multiple sources
• answer
• query languages for XML

10
Querying XML Data

• Traditional databases
• store in a native database
• use a traditional database query language
• i.e. DB4XML
• requires schema representation rigid
• Develop a language for XML
• some inspired by traditional DQLs
• incorporate relational algebra
• others inspired by XML

11
Examples of XML Query Languages

• Lorel (Stanford)
• originally for semi-structured data (Lore)
• extended for XML
• XML-QL (ATT Labs)
• extends SQL
• XQL (Texcel, webMethods, Microsoft)
• for selecting/filtering elements and text of XML
  documents
• natural extension of XSL pattern syntax

12
Features of DQLs in XQLs

• Specific data model
• Lorel XML element is a pair lteid, valuegt
• ltPerson idP1 nameJohn Smith colleagueP2gt
• XML-QL document modeled by XML graph
• node refers to an object identifier
• edges labeled with element tag identifiers
• each graph has a root
• XQL assumes the implied XML data model

13
Features of DQLs in XQLs

• Basic query abstractions
• select supported by all
• joins
• Lorel and XML-QL support joins over the same
  document or several documents
• XQL allows joins only among data in the same
  document
• semantics
• Lorel returns pointers to objects in original
  document
• XML-QL and XQL return new documents that can be
  queried indepentenly

14
Features of DQLs in XQLs

• Path expressions
• a form of navigational queries
• partially specified expressions
• supported by all
• XQL path like a file path
• pse matched with cyclic data
• supported by Lorel and XML-QL
• undefined in XQL

15
Features of DQLs in XQLs

• Quantification, negation and reduction
• existential quantification
• supported by all
• universal quantification
• not supported by XML-QL
• negation
• not supported by XML-QL
• reduction
• supported by none

16
Features of DQLs in XQLs

• Restructuring abstractions
• new elements
• Lorel and XML-QL have functions for construction
• XQL provides no construct mechanism
• grouping
• not in XQL
• Skolem function for OIDs
• could be used for grouping applied to attributes
• not supported by XQL

17
Features of DQLs in XQLs

• Aggregation, nesting and binary queries
• aggregates
• supported by all
• nested queries
• XML-QL and Lorel allow nested queries at any
  level
• unsupported in XQL
• binary queries
• XML-QL does not support difference

18
Features of DQLs in XQLs

• Order management
• ordered results
• XML-QL and Lorel order by given attribute
• preserving order
• (naturally) supported by all
• instance order querying
• only in Lorel and XML-QL

19
Features of DQLs in XQLs

• Typing
• type coercion
• partially supported in XQL
• fully supported in Lorel
• Updates to data
• insert, delete, update
• only in Lorel

20
Features of XQLs

• XML integration
• XML/RDF schema support
• not supported
• cross-document linking
• not supported
• support of tag variables
• in XML-QL variables can be associated with tags
• in Lorel
• obtain all paths reachable from a path expression
• used for building the names of the query result

21
Desired Qualities

• Declarative content of the result defined by the
  query
• Lorel resembles a calculus-based language
• XML-QL represents a logic language
• XQL uses URL-like patterns
• Expressive Power
• Lorel
• XML-QL
• XQL

22
Direction of XML

• Semantic Web
• common framework
• allows data to be shared/reused over application,
  enterprise, and community boundaries
• provides ability to give meaning to data
• based on RDF
• using XML syntax and URIs for names
• more expressive than XML

http//www.w3.org/2001/sw/
23
Data in the Semantic Web

• RDF Resource Description Framework
• standard representation language based on labeled
  graph
• nodes are resources (literals)
• edges are properties
• schema definition language (RDF/S)
• create vocabularies of labels
• XML syntax

http//www.w3.org/RDF/
24
RDF Graph
25
Querying the Semantic Web

• RQL RDF Query Language
• adapts functionalities of semi-structured and
  XML query languages
• for real-scale SW applications
• knowledge portals
• E-market places
• manage voluminous RDF bases and schemas
• currently utilized in many projects

http//www2002.org/CDROM/refereed/329/
26
Querying the Semantic Web

• RQL RDF Query Language
• uniformly navigate/filter RDF graphs
• schema-level
• Instance-level
• highly expressive
• typed functional query language
• relies on a formal graph model
• Interpretation of multiple RDF schemas

http//www2002.org/CDROM/refereed/329/
27
RQL Example
28
Beyond Query Languages

• Query languages may not be enough
• View Languages
• Relational databases
• SQL is query and view language
• XML
• Active Views (INRIA)
• RDF
• RVL RDF View Definition Language

29
References

• Comparative Analysis of Five XML Query Languages
• RQL A Declarative Query Language for RDF
• Viewing the Semantic Web Through RVL Lenses
• World Wide Web Consortium

30
Comments Questions?